libattistella/dev-frontend-ai (agent-observability-pattern)

Conventions for observing a multi-step AI SDK agent in a Next.js app — how to model every agent boundary as a closed TypeScript discriminated union, emit it from a single recorder that isolates volatile experimental_ callback names, stream it as custom data-trace-event parts over the same UI message stream, reduce it in a store, and derive the view through pure selectors with cost computed in one pricing module. Use whenever creating, updating, or porting the src/lib/trace module — trace event union, recorder, pricing, run store, selectors, or the agent run route. Triggers include: 'observe the agent', 'trace events', 'agent timeline', 'token usage', 'tool call timeline', 'stream agent steps', 'cost per run', 'add a trace event', 'instrument the agent'.

Claude Skills are markdown + scripts loaded into Claude Code; scanned with the same pipeline.

Source:

github.com/libattistella/dev-frontend-ai

Provenance:

Found by crawling (provenance, not a security rating)

We found this by crawling public sources and nobody has attested to it. That is how most of the registry gets here, and it is not a mark against the artifact — it only means no authenticated act of publication is on record. An artifact can be the vendor's own and still appear here.